View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

FEBS Letters 588 (2014) 2830–2836

journal homepage: www.FEBSLetters.org

Interaction between NOD2 and CARD9 involves the NOD2 NACHT and the linker region between the NOD2 CARDs and NACHT domain

Rhiannon Parkhouse a, Joseph P. Boyle a, Sophie Mayle a, Kovilen Sawmynaden b, Katrin Rittinger b, ⇑ Tom P. Monie a,c,

a Department of Biochemistry, University of Cambridge, Cambridge, UK b Division of Molecular Structure, MRC-National Institute for Medical Research, London, UK c Department of Veterinary Medicine, University of Cambridge, Cambridge, UK

article info abstract

Article history: NOD2 activation by muramyl dipeptide causes a proinflammatory immune response in which the Received 8 April 2014 adaptor CARD9 works synergistically with NOD2 to drive p38 and c-Jun N-terminal kinase Revised 29 May 2014 (JNK) signalling. To date the nature of the interaction between NOD2 and CARD9 remains undeter- Accepted 6 June 2014 mined. Here we show that this interaction is not mediated by the CARDs of NOD2 and CARD9 as pre- Available online 21 June 2014 viously suggested, but that NOD2 possesses two interaction sites for CARD9; one in the CARD–NACHT Edited by Renee Tsolis linker and one in the NACHT itself.

Structured summary of protein interactions: Keywords: Nucleotide-binding leucine-rich repeat NOD2 physically interacts with CARD9 by anti tag coimmunoprecipitation (View interaction) containing receptor Crohn’s Disease Ó 2014 The Authors. Published by Elsevier B.V. on behalf of the Federation of European Biochemical Caspase activation and recruitment domain Societies. This is an open access article under the CC BY license (http://creativecommons.org/licenses/ Stress kinase pathway by/3.0/). Innate immunity

1. Introduction protein RIP2 (receptor interacting protein 2). Understanding how NOD1/2 signalling is regulated is important for the future develop- The cytoplasmic NOD (nucleotide-binding oligomeri- ment of therapeutic treatments targeting inflammatory disorders sation domain containing) 1 and NOD2 are members of the NLR such as Crohn’s Disease. (nucleotide-binding, leucine-rich repeat containing receptor) fam- CARD9 is an important adaptor molecule in the innate immune ily of pattern recognition receptors. They act as immune sentinels response. CARD9 is predominantly associated with NF-jB signal- and play an important role in combating bacterial infection and ling pathways following stimulation of C-type lectin receptors like maintaining cellular homeostasis [1]. NOD1 and NOD2 recognise DECTIN-1 [10,11]. Receptor ligation upregulates Spleen tyrosine fragments of bacterial peptidoglycan via their C-terminal leucine kinase (SYK) and activates Protein Kinase C delta which phosphor- rich repeats [2–5]. Activation causes conformational change and ylates Thr231 in CARD9 [12]. This causes formation of a CARD9/B relocalisation to the plasma membrane [6–9]. NF-jB (nuclear fac- Cell lymphoma 10 (Bcl-10)/Mucosa-associated lymphoid tissue tor kappa B)-mediated pro-inflammatory signalling pathways are lymphoma translocation protein 1 (Malt1) ‘signalosome’ which activated following interaction between the CARDs (caspase activates NF-jB [10,13]. CARD9 is also involved in NF-jB signalling activation and recruitment domain) of NOD1/2 with their adaptor downstream of the Retinoic acid-inducible I receptor (RIG-1) family [14]. The biological function of CARD9 is conserved between mice and humans. Human CARD9 is able to restore signalling in Abbreviations: NOD, nucleotide oligomerisation domain; NLR, nucleotide-bind- murine Card9 knock-out cells [15] and both inactive human CARD9 ing leucine-rich repeat containing receptor; NF-jB, nuclear factor kappa B; CARD, mutant cells and murine Card9 knockout cells display a defective caspase activation and recruitment domain; RIP2, receptor interacting protein 2; response to b-glucan stimulation [15,16] SNP, single nucleotide polymorphism; MBP, maltose binding protein ⇑ Corresponding author. Address: Department of Veterinary Medicine, University Recently, Hsu and colleagues demonstrated in mice that CARD9 of Cambridge, Madingley Road, Cambridge CB3 0ES, UK. Fax: +44 (0)1223 337610. is required for the synergistic activation of p38 and JNK (c-Jun N- E-mail address: [email protected] (T.P. Monie). terminal kinase) following stimulation of NOD2 by either muramyl

http://dx.doi.org/10.1016/j.febslet.2014.06.035 0014-5793/Ó 2014 The Authors. Published by Elsevier B.V. on behalf of the Federation of European Biochemical Societies. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/). R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836 2831 dipeptide or Listeria monocytogenes [17]. The association between (Abcam) primary antibodies and goat anti-rabbit (Abcam) or goat NOD2 and CARD9 was enhanced by the presence of RIP2 in both anti-mouse (Sigma) secondary antibodies. over-expression and endogenous systems [17]. The relationship between CARD9 and NOD2 is particularly intriguing as the 2.3. Recombinant protein expression and co-sonication pull-down for both these proteins contain polymorphisms influencing suscep- assays tibility to Crohn’s Disease in humans [18,19]. Multiprotein complexes play a key role in innate immune sig- Recombinant proteins were expressed in 10 ml cultures of nalling. Complexes such as the inflammasome and Myddosome Rosetta2 Escherichia coli for 16 hr at 22 °C except for GB1-RIP2– are formed through interactions between members of the death CARD which used 4 h at 37 °C. Cultures were pelleted, frozen over- domain superfamily [20], which includes CARDs. NOD2 and CARD9 night and resuspended in standard lysis buffer (100 mM NaCl, have two and one N-terminal CARDs respectively. We have used 25 mM sodium phosphate, 20 mM imidazole, 5 mM b-mercap- cell-based immunoprecipitation and co-purification of overexpres- toethanol, 0.1 % Triton-X100). Appropriate samples were combined sed recombinant protein to study the molecular details of the and sonicated on ice. Insoluble debris was removed by centrifuga- interaction between NOD2 and CARD9. Unexpectedly, we did not tion (16000Âg,4°C, 10 min) and proteins purified using amylose find any evidence for an interaction between the CARDs of NOD2 affinity chromatography. Eluted samples were visualised by Coo- and CARD9 as previously suggested. Instead, we show that massie Brilliant Blue staining. the region in NOD2 responsible for the interaction with CARD9 involves the NACHT domain and the preceding linker to the 2.4. NMR chemical shift assays CARDs. Samples of unlabelled murine CARD9 CARD and 15N-labelled 2. Materials and methods human NOD2 (28–218) were buffer exchanged overnight into 20 mM sodium phosphate (pH 7.1), 100 mM NaCl and 5 mM DTT. 1 1 15 2.1. Plasmids 1D ( H) and 2D ( H/ N) HSQC NMR spectra were recorded on a Bruker Avance spectrometer at 600 MHz proton frequencies and Full length murine CARD9 (GENBANK: NP_001032836.1) with a processed using an associated software package. All spectra were C-terminal V5-His epitope tag in the pEF6 expression vector (pEF6- recorded at 180 lM sample concentration (with 10% added D2O) mCARD9-V5) was a kind gift from David Underhill [11]. Full-length and at 298 K. human NOD1 (GENBANK: AAD28350.1) and NOD2 (GENBANK: AAG33677.1) with an N-terminal FLAG tag in a pCMV backbone 2.5. Homology modelling and bioinformatics (pCMV-FLAG-NOD1 and pCMV-FLAG-NOD2) were kindly provided by Thomas Kufer [21]. Single nucleotide polymorphisms (SNP) Residues 217–820, corresponding to residues encoded by the across the NACHT domain were identified in the NCBI SNP data- large, central exon of human NOD2, were submitted to the PHYRE2 base and cloned into full-length pCMV-FLAG-NOD2 using site- server for automated modelling [24]. Only one NLR NACHT structure, directed mutagenesis. N-terminal and C-terminal NOD2 constructs the 4KXF structure of NLRC4 [25], is available in the PDB at present were also generated by site-directed mutagenesis. GB1-RIP2–CARD and this was used as a template. Inspection of the resulting align- (human) and the tandem human NOD2 CARD construct used for ment and model showed that while the NBD, HD1 and WH domains NMR chemical shifts have been described previously [22]. The (217–632) were clearly homologous between NOD2 and NLRC4, the CARD of Card9 (residues 6-100), the tandem CARDs of NOD2 rest of the exon, which forms HD2 and part of the LRR domain in (residues 28-215) and the RIP2 CARD (residues 432-534) were NLRC4, showed a more ambiguous alignment and so was not consid- amplified by PCR and inserted, using GatewayÒ cloning, into ered further. The final model represents NOD2 residues A217–C632. the expression plasmid pDEST-HisMBP [23]. In addition to the Homology models of the CARD of murine CARD9 and human N-terminal His6-MBP (maltose binding protein) fusion each CARD9 from residue 6–100 were built using the CARD from construct included a C-terminal FLAG tag to further aid expres- CARD11 (PDB 1D: 4I16) [26] as a template. Models were generated sion and stability. The CARD of murine CARD9 was also inserted using the model-default script from the MODELLER package v9.12 into pDEST-17 to generate an N-terminally 6His tagged (http://salilab.org/modeller/). construct. The amino acid sequences of full length murine CARD9 (GENBANK: NP_001032836.1) and human CARD9 (GENBANK: 2.2. Immunoprecipitation NP_43470.2) were aligned and the percentage identity and similar- ity calculated using Clustal Omega [27]. HEK 293T cells were maintained in DMEM (Sigma) supple- mented with 10% FCS, 100 lg/ml Penicillin/Streptomycin and 3. Results

2mML-glutamine at 37 °C and 5% CO2. Cells were seeded in 6 well plates and transfected using jetPEI (Polyplus Transfection) with 3.1. There is no interaction between the CARDs of NOD2 and CARD9 1 lg/well of full-length, mutated, or truncated pCMV-FLAG- NOD2, or full-length pCMV-FLAG-NOD1, and 1 lg pEF6-mCARD9- CARD-mediated protein–protein interactions play an important V5. After 24 h cells were washed twice in 1 Â PBS and lysed in role in numerous immune signalling pathways such as RIG-1 med- 300 ll RIPA buffer (50 mM Tris–HCl pH 7.6, 150 mM NaCl, 0.25% iated viral sensing, inflammasome formation and facilitating Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate) supplemented NOD1/2-mediated NF-jB signalling via RIP2 [28–30]. NOD2 and with 1 Â Protease Inhibitor Cocktail set V (Calbiochem) and 7.5 CARD9, which both possess CARDs, work synergistically to drive units of Benzonase nuclease (Sigma) per well. Lysates were incu- p38 and JNK signalling following activation of NOD2 and have been bated on ice for 10 min with shaking and clarified by centrifugation shown to interact [17]. Currently nothing is known about how (16000Âg; 2 min; 20 °C). Cells were lysed and incubated with Pro- this interaction is mediated, although theoretical models have tein G coated magnetic Dynabeads (Life Technologies) coupled to suggested the involvement of CARD:CARD interactions [31]. mouse anti-FLAG antibody (Sigma). Immunoprecipitated proteins To test whether the NOD2 CARD9 interaction is CARD-mediated and inputs were detected by western blot analysis using: rabbit we expressed the CARDs of human NOD2, murine CARD9 and human anti-flag (Sigma), mouse anti-V5 (Abcam) and mouse anti-GAPDH RIP2 as recombinant proteins fused to solubility-enhancement and 2832 R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836 epitope tags. Interactions were assessed using a modified version interaction between the CARD of CARD9 and either the NOD2 of the protocol established to study interactions between the tandem CARDs or the RIP2 CARD; although the CARD9 CARD could CARDs of NOD2 and RIP2 [22]. Specifically, bacterial cultures facilitate weak self-association as has been previously reported containing overexpressed recombinant protein were resuspended [32] (Fig. 1A and B). The inability of the tandem CARDs of NOD2 and combined prior to sonication. Amylose resin based affinity and the CARD9 CARD to interact was further confirmed using chromatography was used to purify proteins fused to MBP and NMR chemical shift studies. 1D (1H) NMR spectrum confirmed co-purifying interacting proteins were detected by Coomassie that the CARD of murine CARD9 is tertiary structured as shown Brilliant Blue staining following SDS–PAGE. by the ring-current shifted methyl signals (<0.5 ppm) resulting Consistent with previous work the NOD2 CARDs successfully from formation of a hydrophobic core (Fig. 1C). 15N-labelled pulled down the RIP2 CARD, which was also capable of mediating NOD2 (28–218) is also tertiary structured, displaying a generally homomeric RIP2 interactions (Fig. 1A). However, we detected no good dispersion of peaks in the 2D (1H/15N) HSQC spectra and

A B HM-NOD2 CARDs + + -- -- HM-NOD2 CARDs -- --+ + HM-CARD9 CARD -- + + -- HM-CARD9 CARD -- + + - - HM-RIP2 CARD -- -- + + His-CARD9 CARD + + + + + + GB1-RIP2 CARD + + + + + +

75 75 50 50 37 37 25 25 20 * 20 15 * 15

TETETE T E TETE C D

E

Fig. 1. The CARDs of NOD2 and CARD9 do not interact. His-MBP tagged NOD2 CARDs, CARD9 CARD and RIP2 CARD were used to co-purify GB1-RIP2 CARD (A) or His-CARD9 CARD (B). The CARD9 CARD did not interact with either the CARDs of NOD2 or RIP2, but did display homomeric interactions. The asterisks represent the location of the GB1- RIP2 CARD (A) and the His-CARD9 CARD (B). T = total cell lysate; E = Elution post amlyose affinity purification. (C) 1D (1H) NMR spectrum of CARD9 confirms the protein is tertiary structured. (D) Overlay of 2D (1H/15N) HSQC spectra of NOD2 (28–218) in the absence (red; 1:0 equivalents) and presence (black; 1:1 equivalents) of CARD9 CARD. (E) Side-by-side 2D (1H/15N) HSQC spectra of NOD2 (28–218) in the absence (left) and presence (right; 1:1 equivalents) of CARD9 CARD. R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836 2833 LRR A + NACHT- -NOD2 + NOD2 + NOD1 + CARD + CARD-NACHT + kDa + LRR V5-CARD9 Flag Co-IP V5 72 CARD9

130 90 72 Flag NOD2 55 domains

Input 36

V5 72 CARD9 55 GAPDH 36 B E441K D379A L550V R334W W355stop A612V + NOD2 + + + S431L + + + R702W + I363F + L248R + + D357A + + P463A kDa + V5-CARD9 Flag 72 Co-IP V5 CARD9

130 90 Flag 72 NOD2 55 domains 36 Input

V5 72 CARD9 GAPDH55 36 W355Stop S431L L248R R334W I363F, D379F E441K, P463A L550V A612V R702W

227NACHT 744 D357A C S431 E441 P463 L248 180 A612 D379

N L550 R334 I363

D357

Fig. 2. NOD2 interacts with CARD9. (A) HEK293T cells were transiently transfected with V5-CARD9 and FLAG-NOD2 full-length and domain truncation expression constructs; or with (B) V5-CARD9 and FLAG-NOD2 NACHT polymorphism containing constructs. 24 h later cell lysates were subjected to co-immunoprecipitation using anti-FLAG antibody and samples analysed by Western-blotting. Neither the CARDs alone, nor the LRRs alone interacted with CARD9; and none of the polymorphisms disrupted the interaction. The relative position of the polymorphisms is shown on a schematic of NOD2. (C) Location of the NOD2 SNPs on a homology model of the NOD2 NACHT. SNPs are coloured red and the side chains shown as spheres. Images were generated using PYMOL (Schrödinger). chemical shift values >9 ppm (due to hydrogen bond formation; these approaches conclusively support the view that the CARDs of Fig. 1D and E). However, there are no significant chemical shift NOD2 and CARD9 do not directly interact. changes in the presence of equimolar CARD9 CARD indicating that Comparing the sequences of murine CARD9 and human there is no interaction between the two recombinant proteins CARD9 indicated that these proteins are 86% identical and 91% under the conditions studied (Fig. 1D and E (right panel)). Together similar across their entire sequence, both identity and similarity 2834 R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836

A

CARD1 CARD2227 NACHT 744 LRR 1040 NOD2 NACHT-LRR CARD-NACHT CARDs 355 W355stop 335 Q335stop 315 L315stop 274 A274stop 260 W260stop 240 Y240stop 227 Start T227 254 Start E254 274 Start A274

B L315stop + W355stop + start E254 + + NOD2 + + CARDs + A274stop + + start A274 + W260stop + Y240stop + Q335stop kDa V5-CARD9 + NOD2 + start T227 kDa Flag V5 72 V5 72 Co-IP CARD9

130 130 90 90 72 72 Flag Flag

55 NOD2 55 36 36 Input

V5 72 V5 72 CARD9 55 55 GAPDH 36 36

Fig. 3. (A) Schematic representation of the NOD2 deletion constructs. (B) HEK293T cells were transiently transfected with truncated FLAG-NOD2 expression constructs and V5-CARD9 and co-immunoprecipitated after 24 h using anti-FLAG antibody and samples analysed by Western-blotting. Inclusion of the NOD2 CARD–NACHT linker, or the NACHT domain itself, facilitated interaction with CARD9.

increasing to 95% for the CARD (Supplementary figure 1A). Homol- NACHT on the interaction with CARD9. This included the widely ogy models of the two CARDs confirmed that four of the five studied R702W SNP. We also tested the hyperactive Blau Syn- substitutions were in helix 1 and the fifth in helix 6. None of the drome associated SNP R334W [37]. All of the polymorphisms still mutations affect the overall protein fold or the electrostatic prop- interacted with CARD9 indicating that these residues, and poten- erties of the CARDs (Supplementary Fig. 1B and C). This is consis- tially the surrounding regions of the NACHT, were not crucial for tent with the conserved biological function of the molecules and CARD9 interaction (Fig. 2B and C). It also demonstrates that the allows us to conclude that it is highly unlikely that the CARDs of clinical impact of the R334W and R702W SNPs is unlikely to result human CARD9 and human NOD2 would interact either. from alteration of the CARD9 and NOD2 interaction. Interestingly, the W355stop SNP, which lacks any sequence downstream of this 3.2. The NACHT and CARD–NACHT linker of NOD2 are important tryptophan, still immunoprecipitated Card9 suggesting that a for interaction with CARD9 region between the NOD2 CARDs and W355 could mediate interac- tion. Consequently, we generated further C-terminal NOD2 trunca- Given the inability of the respective CARDs to interact we used tions (Fig. 3A) and tested their interaction with CARD9. All of these domain truncations of FLAG-tagged NOD2 to map the region truncations, including A274stop which lacks any of the NACHT, required for interaction with CARD9. HEK293 cells were transiently retained the ability to interact with CARD9 (Fig. 3B). Together with transfected with NOD2 truncations and V5-tagged CARD9 and the failure of the CARD only construct (residues 1–227; Figs. 2A proteins immunoprecipitated after 24 h. CARD9 was immunopre- and 3B) to interact with CARD9 this indicated that residues 228– cipitated by full-length, CARD–NACHT, and NACHT–LRR NOD2 274 contain a critical region for CARD9 interaction. constructs (Fig. 2A). Neither the CARDs nor the LRRs alone inter- To characterise the role of the CARD–NACHT linker in more acted with CARD9, suggesting that the NACHT domain of NOD2 detail we expanded our range of mutants to alter the amount of has a critical role in mediating interaction with CARD9. linker present (Fig. 3A). Extension of the C-terminus of the CARD To further investigate the importance of the NOD2 NACHT only construct to residue 240 did not result in interaction with domain we tested the impact of a panel of Crohn’s Disease associ- CARD9 (Fig. 3B). However, addition of a further twenty residues, ated single nucleotide polymorphisms (SNP) [33–36] spanning the to position 260, resulted in CARD9 being immunoprecipitated, R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836 2835 albeit more weakly than with the the wild-type protein, or other In conclusion, we have identified regions of NOD2 important for truncations (Fig. 3B). All of our N-terminal truncations were able CARD9 interaction. These now require functional interrogation to to immunoprecipitate CARD9, including a construct beginning at determine precisely how the two proteins interact and whether 274 which therefore lacks the CARD–NACHT linker (Fig. 3B). these regions could serve as modulators of receptor signalling with Intriguingly both the constructs 1–274 and 274–1040 interacted the potential for the regulation of a range of signalling pathways. with Card9 thereby implying the presence of at least two binding sites in NOD2 for CARD9; one located in the CARD–NACHT linker Acknowledgements (between residues 241 and 274) and the other in the NACHT domain. This work was funded by a Wellcome Trust Career Develop- Together these results support an interaction between NOD2 ment Fellowship (WT085090MA) to TPM and a Medical Research and CARD9. This interaction is not mediated by the CARD domains, Council grant (U117565398) to KR. RP and JPB were supported but instead involves two regions in NOD2 – one linking the CARDs by BBSRC Doctoral Training Grants. None of the funders had any and NACHT, the other within the NACHT itself. role in the design of the study. We thank Geoff Kelly for assistance with the acquisition of NMR spectrum.

4. Discussion Appendix A. Supplementary data

The pattern recognition receptors NOD1 and NOD2 play an Supplementary data associated with this article can be found, in important role in the pro-inflammatory immune response to bac- the online version, at http://dx.doi.org/10.1016/j.febslet.2014.06. terial infections. Their ability to activate NF-jB-mediated tran- 035. scription following interaction with RIP2 has been widely studied. Our understanding of how these receptors activate alter- References native signalling pathways, such as those utilising p38 and JNK, is limited. However, it has been previously shown in mice that [1] Philpott, D.J. et al. (2014) NOD proteins: regulators of inflammation in health engagement of the adaptor protein CARD9 is crucial in facilitating and disease. Nat. Rev. Immunol. 14, 9–23. NOD2-initiated p38 and JNK signalling, with the two proteins [2] Monie, T.P. (2013) NLR activation takes a direct route. Trends Biochem. Sci. 38, 131–139. working in synergy to mediate this response [17]. How this is [3] Mo, J. et al. (2012) Pathogen sensing by nucleotide-binding oligomerization achieved, and how NOD2 and CARD9 interact has not previously domain-containing protein 2 (NOD2) is mediated by direct binding to been understood. In this work we have identified two regions of muramyl dipeptide and ATP. J. Biol. Chem. 287, 23057–23067. [4] Grimes, C.L. et al. (2012) The innate immune protein Nod2 binds directly to NOD2 capable of binding CARD9; one in the linker connecting MDP, a bacterial cell wall fragment. J. Am. Chem. Soc. 134, 13535–13537. the CARDs and the NACHT, and one in the NACHT itself. These [5] Laroui, H. et al. (2011) L-Ala-c-D-Glu-meso-diaminopimelic acid (DAP) binding surfaces may by spatially adjacent to one another in the interacts directly with leucine-rich region domain of nucleotide-binding oligomerization domain 1, increasing phosphorylation activity of receptor- tertiary structure of the protein. A random mutagenesis study of interacting serine/threonine-protein kinase 2 and its interaction with NOD2 identified two residues, A232 and V256, in the linker region nucleotide-binding oligomerization domain containing 1. J. Biol. Chem. 286, that when mutated resulted in a significant loss of NF-jB signalling 31003–31013. [6] Barnich, N. et al. (2005) Membrane recruitment of NOD2 in intestinal [38]. Whilst for A232 this correlated with a loss in protein expres- epithelial cells is essential for nuclear factor-{kappa}B activation in muramyl sion this was not the case for V256, suggesting that the linker may dipeptide recognition. J. Cell Biol. 170, 21–26. be more important for NOD2 functionality than previously [7] Kufer, T.A. et al. (2008) The pattern-recognition molecule Nod1 is localized at the plasma membrane at sites of bacterial interaction. Cell. Microbiol. 10, 477– assumed. Somewhat surprisingly and despite the known impor- 486. tance of CARDs in mediating protein–protein interaction in [8] McDonald, C. et al. (2005) A role for Erbin in the regulation of Nod2-dependent immune signalling complexes we saw no evidence that the CARDs NF-kappaB signaling. J. Biol. Chem. 280, 40301–40309. of NOD2 and CARD9 could interact with one another. [9] Lécine, P. et al. (2007) The NOD2–RICK complex signals from the plasma membrane. J. Biol. Chem. 282, 15197–15207. In this study we used murine CARD9 and human NOD2. In light [10] Gross, O. et al. (2006) Card9 controls a non-TLR signalling pathway for innate of the conserved biological function [15,16]; the high level of anti-fungal immunity. Nature 442, 651–656. sequence identity (Supplementary Fig. 1A); and the conserved nat- [11] Goodridge, H.S. et al. (2009) Differential use of CARD9 by dectin-1 in macrophages and dendritic cells. J. Immunol. 182, 1146–1154. ure of the electrostatic surfaces (Supplementary Fig. 1C) between [12] Strasser, D. et al. (2012) Syk kinase-coupled C-type lectin receptors engage the human and murine proteins we are confident that our observa- protein kinase C-r to elicit Card9 adaptor-mediated innate immunity. tions and conclusions remain valid for interactions between Immunity 36, 32–42. [13] Hara, H. et al. (2007) The adaptor protein CARD9 is essential for the activation human CARD9 and NOD2. of myeloid cells through ITAM-associated and Toll-like receptors. Nat. Our observations raise interesting questions about the molecular Immunol. 8, 619–629. composition and formation of the NOD2:CARD9 signalling complex. [14] Poeck, H. et al. (2010) Recognition of RNA virus by RIG-I results in activation of CARD9 and inflammasome signaling for interleukin 1 beta production. Nat. The lack of NOD2 CARD9 CARD-mediated interaction suggests that Immunol. 11, 63–69. NOD2 could interact concurrently with RIP2, via its CARDs, and with [15] Glocker, E.-O. et al. (2009) A homozygous CARD9 mutation in a family with CARD9 via its linker/NACHT. This would enable simultaneous activa- susceptibility to fungal infections. N. Engl. J. Med. 361, 1727–1735. [16] LeibundGut-Landmann, S. et al. (2007) Syk- and CARD9-dependent coupling tion of NF-jB, p38 and JNK signalling pathways from the same mac- of innate immunity to the induction of T helper cells that produce interleukin romolecular complex. Hsu and colleagues reported that the presence 17. Nat. Immunol. 8, 630–638. of RIP2 enhances the interaction between NOD2 and CARD9 [17] Hsu, Y.-M.S. et al. (2007) The adaptor protein CARD9 is required for innate [17].This could be through promotion of conformational changes immune responses to intracellular pathogens. Nat. Immunol. 8, 198–205. [18] Zhernakova, A. et al. (2008) Genetic analysis of innate immunity in Crohn’s due to receptor activation which expose the CARD9 binding motifs disease and ulcerative colitis identifies two susceptibility loci harboring on NOD2; or by promoting protein clustering and thereby increasing CARD9 and IL18RAP. Am. J. Hum. Genet. 82, 1202–1210. the avidity of the NOD2:CARD9 interaction. CARD9 contains a pre- [19] Ogura, Y. et al. (2001) A frameshift mutation in NOD2 associated with susceptibility to Crohn’s disease. Nature 411, 603–606. dicted coiled-coil domain (CCD). CCDs are widely reported to medi- [20] Kersse, K. et al. (2011) The death-fold superfamily of homotypic interaction ate protein–protein interactions and it remains entirely feasible that motifs. Trends Biochem. Sci. 36, 541–552. the CCD mediates interaction with NOD2. However, difficulties in [21] Kufer, T.A. et al. (2006) Role for erbin in bacterial activation of Nod2. Infect. Immun. 74, 3115–3124. expressing the CCD independently of the CARD [32] have made this [22] Fridh, V. and Rittinger, K. (2012) The tandem CARDs of NOD2: intramolecular hypothesis impractical to test to date. interactions and recognition of RIP2. PLoS ONE 7, e34375. 2836 R. Parkhouse et al. / FEBS Letters 588 (2014) 2830–2836

[23] Nallamsetty, S. et al. (2005) Gateway vectors for the production of [31] Roth, S. and Ruland, J. (2013) Caspase recruitment domain-containing protein combinatorially-tagged His6-MBP fusion proteins in the cytoplasm and 9 signaling in innate immunity and inflammation. Trends Immunol. 34, 243–250. periplasm of Escherichia coli. Protein Sci. 14, 2964–2971. [32] Bertin, J. et al. (2000) CARD9 is a novel caspase recruitment domain- [24] Kelley, L.A. and Sternberg, M.J.E. (2009) prediction on the containing protein that interacts with BCL10/CLAP and activates NF-kappa Web: a case study using the Phyre server. Nat. Protoc. 4, 363–371. B. J. Biol. Chem. 275, 41082–41086. [25] Hu, Z. et al. (2013) Crystal structure of NLRC4 reveals its autoinhibition [33] Lesage, S. et al. (2002) CARD15/NOD2 mutational analysis and genotype- mechanism. Science 341, 172–175. phenotype correlation in 612 patients with inflammatory bowel disease. Am. [26] Li, S. et al. (2012) Structural insights into the assembly of CARMA1 and BCL10. J. Hum. Genet. 70, 845–857. PLoS ONE 7, e42775. [34] Tukel, T. et al. (2004) Crohn disease: frequency and nature of CARD15 [27] Sievers, F. et al. (2011) Fast, scalable generation of high-quality mutations in Ashkenazi and Sephardi/Oriental Jewish families. Am. J. Hum. protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. Genet. 74, 623–636. 7, 539. [35] Lappalainen, M. et al. (2008) Novel CARD15/NOD2 mutations in Finnish [28] Kobayashi, K. et al. (2002) RICK/Rip2/CARDIAK mediates signalling for patients with Crohn’s disease and their relation to phenotypic variation receptors of the innate and adaptive immune systems. Nature 416, in vitro and in vivo. Inflamm. Bowel Dis. 14, 176–185. 194–199. [36] Schnitzler, F. et al. (2006) Eight novel CARD15 variants detected by DNA [29] Martinon, F. et al. (2002) The inflammasome: a molecular platform triggering sequence analysis of the CARD15 gene in 111 patients with inflammatory activation of inflammatory caspases and processing of proIL-beta. Mol. Cell 10, bowel disease. Immunogenetics 58, 99–106. 417–426. [37] Miceli-Richard, C. et al. (2001) CARD15 mutations in Blau syndrome. Nat. [30] Yoneyama, M. et al. (2004) The RNA helicase RIG-I has an essential function in Genet. 29, 19–20. double-stranded RNA-induced innate antiviral responses. Nat. Immunol. 5, [38] Tanabe, T. et al. (2004) Regulatory regions and critical residues of NOD2 730–737. involved in muramyl dipeptide recognition. EMBO J. 23, 1587–1597.